Can AI Put The UK Back On The Map For Automotive Technology?

Can the UK regain its automotive mojo with the application of advanced AI technology?

getty

The UK used to lead the world in automotive innovation and manufacturing. In the 1950s, the country boasted 52% of the global car export market. By 2024, that had dropped to just 3.75%. With the rise of China, Britain could probably never compete again as a volume auto producer, but could its prowess in AI provide an alternative route to a leading role again? I asked key UK innovators about the possibilities, with Imperial College London at their core.

Monolith AI, Poster Child Of UK Startup Success

One of the highest profile names working in this space is Monolith AI. US-based AI hyperscalar CoreWeave purchased the company in October 2025 for an undisclosed sum, although it is thought to be in the range of hundreds of millions of US dollars. Founder & CEO of Monolith AI, Richard Ahlfeld, reckons the UK has unique characteristics for the AI industry. “The first one is UK university education,” he says, arguing that there is a greater focus on probabilistic systems than other countries, even with disciplines like engineering. “In Germany, probability has nothing to do with engineering. It either works or it doesn’t.” Ahlfeld argues that probabilistic thinking is how AI operates, making it easier for UK-educated people to adapt.

Ahlfeld also cites the proximity to London of companies willing to try out new ideas from people with a credible background, such as Imperial postgraduates. He argues that having McLaren and other F1 teams in the area provides a rich opportunity for innovation, and Rolls Royce is also very open to new possibilities. “That’s how Monolith landed its first clients in the UK,” he says. “I have no idea whether that would have worked anywhere else. It’s not just Monolith. There are other startups in this sector like PhysicsX.”

Imperial college is a key factor. The university has produced innumerable startups, thanks to been fostering an entrepreneurial spirit for over a decade. The CoreWeave acquisition underlines how successful Monolith AI has been with its innovations, which start by hiring automotive engineers and training them in AI. This has enabled the company to provide services for the auto industry that will slot in easily. The best example has been work with Nissan on the new Leaf, which saved the company 17% in testing costs.

Monolith’s original mission since it was founded in 2018 has been “to empower engineers to use AI by themselves,” explains Ahlfeld. While there are lots of AI tools around, most are very hard to use. The core Monolith AI product is providing this “last mile” to non-AI experts. “This is now becoming part of CoreWeave. They’ve acquired some of the top AI platforms in the field from San Francisco and Silicon Valley, including Weights & Biases and Marimo. The thing that these AI platforms don’t have is the last-mile delivery to an engineer in a lab.”

With just 50 engineers based in London, Monolith AI had focused primarily on optimizing batteries. But the CoreWeave acquisition will enable the company to dramatically expand its vision. “Now, we have more than 100X the resources,” says Ahlfeld. “We’re part of a multi-billion-dollar company. From a funding perspective, it’s a complete game changer.” One area Monolith plans to expand back into is aviation, after accouncing a partnership with Vertical Aerospace to work on eVTOL technology. Another will be optimizing the production of silicon for AI accelerators. Then, unsurprisingly, will be more work with F1, including the Aston Martin team, which CoreWeave already sponsors, although Ahlfeld wouldn’t provide any details about this.

However, the 17% test cost reduction provided to Nissan shows what Monolith can do, with a strong UK focus. “It’s a major Japanese OEM,” says Ahlfeld. “It’s rare to get press releases from them unless they’re very happy at a senior level. Nissan has a big R&D center in the UK where they do a lot of testing and quality assurance. We’ve been working with them for quite a while. Now we have a 17% improvement and we want to push this to 30% as quickly as we can.”

This shows how British AI could provide services with huge impact for large global businesses. “If you look at some of the companies that have emerged in AI and engineering, PhysicsX as a Monolith competitor is already valued at a billion now” says Ahlfeld. “The kind of value that is being created through Monolith and other companies is substantial. As software is getting eaten by AI, consultative services are one of the main areas that remains, and that’s something the UK has generally been good at.” Ahlfeld sees robotic lab testing, with experiments defined by AI, as the next big growth area for AI in 2026. This could potentially accelerate product development a thousandfold, and this will be a core focus for Monolith AI with its massive CoreWeave backing.

Polaron Uses AI To Improve Battery Performance

At a lower financial level to Monolith, but still with an impressive trajectory, is Polaron. The company has just raised $8 million investment from Racine2, Speedinvest and Futurepresent, and is another offspring from Imperial College London. “The company comes out of eight years of research at Imperial where I did my PhD with my co-founder,” says Isaac Squires, CEO and Co-Founder, Polaron. “We were exploring the idea of how we can apply AI to accelerate the way that we characterize and design materials with a focus on battery materials, to improve their performance.”

Polaron aims to understand how the industrial manufacturing process affects results by using material microscopy. “In a battery, we can control things like how hard we press it, what the temperatures are that we dry it at, and that fundamentally changes things like how much energy we can store and how fast we can charge the material,” says Squires. “We are trying to understand that relationship between process and performance, using AI to connect those things together, and images of the material at the micro scale. This helps companies to rapidly accelerate their process optimization and design and to unlock higher performing materials.”

The material images are captured at a resolution of a billionth of a meter. Polaron uses a series of these image datapoints to predict how manufacturing process variations might affect performance. “It’s a generative model where the input to the model is the process and the output is a prediction of what the structure will look like and then how that will behave,” says Squires. “It’s predicting a 3D structure that’s changing as we change the recipe of how we make the thing. Then we optimize the recipe to give us the best structure for the best performance.”

The number of data points required varies. It could be as few as 30, or it could be a lot more for a more complex problem across a bigger space. “The key thing is that we’re able to be very data efficient because what you see under a microscope is incredibly rich with information,” says Squires. “If you take an image of a material, you have billions and billions of pixels that are telling you information about the way the material works, which is what allows us to learn more from a small amount of data.” Computer vision, a breed of AI that recognizes patterns, generates data from images.

“We were doing an optimization of a battery material where we were looking at optimizing the way that you mix the slurry,” explains Squires, citing an example of how Polaron’s technology works in practice. “When you make battery material it is like cooking. You make this dough initially and then you cook that onto a foil and then you dry it. Then you crush it down (called calendaring) and then you roll it up and you put it in a can. We have control over each of these different steps – how hard you mix it, how thick you coat it, how hard you crush it. All these things can be optimized and that changes the structure. Every time we change those parameters, the structure changes and that alters the performance.” This includes energy density and charging speed.

“The model predicted that the optimal calendaring density to achieve high performance was a lot denser than the company’s engineers expected,” continues Squires. The results were significant. “We ended up making the cell about 11% higher energy density,” says Squires. “The secret sauce is in proprietary generative models that we’ve developed. If you can improve the performance of battery electrodes by 10-20%, this is billions of dollars of value creation for companies, but the models that we’re developing don’t cost billions of dollars to train.”

The big saving is in how rapidly new material variations can be found with potential benefits, without needing to perform so many expensive physical tests. “You can explore the design space 10,000 times faster because the ability to make a prediction and test the performance of that prediction inside these machine learning models takes about 5 to 10 seconds whereas making these things in reality takes a lot longer than that.” With one of Polaron’s clients, the company is saving thousands of engineering hours by providing an automated AI material analysis workflow that takes under an hour. “The most fundamental value proposition that we have to our customers today is that there are things that they simply cannot do without AI.”

As Polaron grows, it will begin to look beyond batteries to other materials, but this is currently the company’s key focus. “Our customers touch around a third of global EV manufacturing. We look at dendrite growth, swelling in high silicon content anodes. We’re looking at the cracking mechanisms in cathodes and anodes. We’re looking at delamination in solid state batteries. We’re working on novel new materials for cathodes and anodes, emerging high nickel content and lithium manganese iron phosphate (LMFP) batteries.” LMFP is a development of LFP, with greater density but the same robustness. “We’re talking to some of the largest global EV manufacturers in the world.”

“AI is the new industrial revolution,” concludes Squires. “No matter what your job is, no matter what sector you work in, everybody is asking what AI means for them.”

Imperial College London’s Central Role In The UK AI Revolution

Both Monolith AI and Polaron came from Imperial College London, as was Breathe Battery Technologies, which now works with Volvo. Aldo Faisal, Professor of AI and Neuroscience at Imperial, explains why the university plays such a key role in the application of AI to manufacturing. “Being an Imperial spin out already nets you the first million in terms of venture capital funding,” he says. “That’s the reputation of this institution.”

While Faisal’s focus is on the medical application of AI. AI has had huge impact reducing times for drug discovery, particularly for treating rare diseases. However, he sees the holistic application of this technology across fields as a key area for development. “You really need to do something that we call convergence science,” he says. “We’re interested in thinking through what you need to have an impact on the world to tackle some of the world’s biggest problems – sustainability, battery, and self-driving, alongside healthcare.”

To this end, Imperial has formed a School of Convergence Science to think through these big-scale problems. Both Monolith AI and Polaron came from within this School. Faisal’s own project is Nightingale AI, which aims to become the world’s largest biomedical and medical AI foundation model. The UK has strong potential in this area partly because of its NHS, which integrates all medical services into one system, where countries like the USA are more fragmented, and patients fall out of the data pool when they lose insurance coverage.

Faisal talks of the “large everything model”, which ingests a far greater range of data than just text or images. “Imperial is a hotbed for this type of AI of science activity because we’re the number one institution for science and research in Europe, second to none,” agues Faisal. “Depending on the world rankings that you’re looking at, we’re number two, number five, or number four in the world. AI at Imperial started in 1956, just after the Dartmouth conference where the claim of building a universal problem solver was made. We now have over 250 AI professors at Imperial.”

An example of converged science is a project Faisal did in 2020 with former Formula E champion Lucas di Grassi, now a driver for Lola Cars. “We wanted to know how we can exploit his human intuition when a car loses control,” says Faisal. “We rented the Top Gear racetrack, got him a supercar, put sensors all over it, turned it into a living lab, and then got him in freezing conditions to drive never under 60 miles per hour without the assistive systems. We hooked up to the car electronics to get all sorts of information. We hooked him also up to brain sensors and eye tracking and movement sensors. We could exactly predict when he was expecting the car to lose control and when the car actually lost control. It was here that the AI systems or the car’s electronic systems were not so good at predicting it as he was. We’d shown how you can train self-driving cars in extreme situations much faster. You might not need a million kilometers, maybe just 700,000- or 800,000-kilometers driving experience for the AI. They also reach much better levels of performance overall.”

“Imperial is punching far above its weight in AI,” concludes Faisal. “We are the number three country for AI after the US and China. In terms of AI scale-ups and AI company formations, London outpaces San Francisco.” Faisal argues that it’s essential that we develop AI. “We’re at risk of having AI systems that we can’t control or that somebody else can just block for trade dispute reasons. We don’t need a British ChatGPT. There are plenty of excellent English language models. But we should build things where we have natural strength. That’s why we have an opportunity here to become world leaders by focusing on AI and science – material science, neuroscience, automotive. London is the leading place for these things to happen, with Imperial as the focal point. We have companies coming out of Imperial that tackle the domain of sustainability and making better cars and industrial products with this integration of science. If we put our ingenuity in the right places, we can win. We don’t have to mass produce low quality things.”